ADAPTIVE DELTA MODULATION

ADAPTIVE DELTAMODULATIONPREPARATION. 142background . 142TIMS adaptive delta modulator. 142the voltage controlled amplifier - VCA.143EXPERIMENT . 144the adaptive control voltage .144VCA calibration . 144manual control.145stability .145adaptive control.145demodulation. 146TUTORIAL QUESTIONS . 147APPENDIX. 148loop stability. 148Adaptive delta modulationVol D1, ch 15, rev 1.0- 141

ADAPTIVE DELTA MODULATIONACHIEVEMENTS: introduction to a variation of the basic delta modulator, whichadjusts the step size according to the slope of the signal being sampledPREREQUISITES: completion of the experiments entitled Delta modulation andDelta demodulation in this Volume.ADVANCED MODULES: DELTA MODULATION UTILITIES; DELTADEMOD UTILITIES; WIDEBAND TRUE RMS METER optional.PREPARATIONbackgroundIt is assumed that you have been introduced to the principles of adaptive deltamodulation in your course work.TIMS adaptive delta modulatorThe basic delta modulator was studied in the experiment entitled Delta modulation.It is implemented by the arrangement shown in block diagram form in Figure 1. messageinLIMITER-SAMPLER VdeltamodulatedsignaloutputINTEGRATORclockkFigure 1: basic delta modulatorYou will remember that with this modulator there was a conflict when determiningthe step size.A large step size was required when sampling those parts of the input waveform ofsteep slope. But a large step size worsened the granularity of the sampled signalwhen the waveform being sampled was changing slowly. A small step size ispreferred in regions where the message has a small slope.142 - D1Adaptive delta modulation

This suggests the need for a controllable step size - the control being sensitive to theslope of the sampled signal. This can be implemented by an arrangement such as isillustrated in Figure 2.message inSAMPLERLIMITER e 2: an adaptive delta modulatorThe gain of the amplifier is adjusted in response to a control voltage from theSAMPLER, which signals the onset of slope overload.The step size is proportional to the amplifier gain. This was observed in an earlierexperiment.Slope overload is indicated by a succession of output pulses of the same sign.The TIMS SAMPLER monitors the delta modulated signal, and signals when there isno change of polarity over 3 or more successive samples.The actual ADAPTIVE CONTROL signal is 2 volt under ‘normal’ conditions, and risesto 4 volt when slope overload is detected.The gain of the amplifier, and hence the step size, is made proportional to thiscontrol voltage. Provided the slope overload was only moderate the approximationwill ‘catch up’ with the wave being sampled. The gain will then return to normaluntil the sampler again falls behind.Much work has been done by researchers in this area, and sophisticated algorithmshave been developed which offer significant improvements over the simple system tobe examined in this experiment.the voltage controlled amplifier - VCAThe VCA can be modelled with a MULTIPLIER. This is shown in Figure 3.inputy(t)k V y(t) outputVk multiplier constantFigure 3: the voltage controlled amplifierThe control in Figure 3 is shown as a DC voltage. This may be set to any value inthe range Vmax. Beyond Vmax. the MULTIPLIER will overload. However, thecontrol voltage need not be DC, but can be time varying. Under these conditions thearrangement is more likely be called a modulator.Adaptive delta modulationD1- 143

You have met the MULTIPLIER constant, ‘k’, in earlier experiments of Part I,where it was defined and measured.EXPERIMENTThe block diagram of Figure 1 was modelled in the experiment entitled Deltamodulation. Refer to that experiment for details.The adaptive delta modulator of Figure 2 differs only by the addition of a voltagecontrolled amplifier (VCA), modelled, as described above, with a MULTIPLIER.the adaptive control voltageThe DELTA MODULATION UTILITIES module has a socket labelled ADAPTIVEOUTPUT. The signal from this socket is at a level of either 2 or 4 volts. The loweroutput is what might be called the ‘normal’ level. If at any time the delta modulatedsignal contains three or more consecutive samples of the same size then this signalgoes to the higher ( 4) volt level 1. Three or more consecutive samples of the samelevel indicates slope overload.When including the VCA in the feedback path you must ensure that at no time willeither of the inputs to the MULTIPLIER exceed its safe (ie, linear) operating range(say 5 volts absolute maximum).VCA calibrationBefore setting up the delta modulator, it is wise to familiarise yourself with theoperation of the VCA.T1 set up a VCA according to the block diagram of Figure 3. Use the VARIABLEDC module as a control signal, and a sinewave as input. Connecteach via a BUFFER amplifier so that the values of Vmax/x and Vmax/ycan be determined. These are the overload levels for each of theinputs. They are likely to be similar. Select DC coupling with thefront panel switch of the MULTIPLIER.T2 measure the VCA gain for a control voltage of 2 volt. This is the ‘normal’output from the ADAPTIVE CONTROL of the DELTA MODULATORUTILITIES module.Your measurements should have shown that the MULTIPLIER can accept inputsconsiderably in excess of the TIMS ANALOG REFERENCE LEVEL beforeoverload sets in.1 more details in the TIMS Advanced Modules User Manual.144 - D1Adaptive delta modulation

Likewise, the INTEGRATOR input can, under some conditions, be subject to quitelarge input signals; but it is robust and can also handle input amplitudes well inexcess of the TIMS ANALOG REFERENCE LEVEL.You will notice that, except for the presence of the MULTIPLIER in the feedbackloop, the modulator is the same as that studied in the experiment entitled Deltamodulation. You should use the same setting up procedure as in that experiment,with the adaptive control inhibited. This is done by connecting 2 volt in place ofthe ADAPTIVE CONTROL voltage to the MULTIPLIER.manual controlT3 model the block diagram of Figure 1. This is not the adaptive modulator.Refer to the experiment entitled Delta modulation for details. Theamplifier in the feedback loop is modelled with two BUFFERamplifiers in cascade.T4 observe the two inputs to the SUMMER. Adjust the feedback gain so that thesawtooth shows some evidence of (ie, moderate) slope overload.T5 observe the control voltage from the ADAPTIVE CONTROL output socket. It willbe alternating between V1 (no slope overload) and V2 volt (followingthe onset of, and coincident with, the slope overload). Record thevalue of V1 (about 2 volt).T6 insert the VCA between the SAMPLER and the BUFFER AMPLIFIER. Set thecontrol voltage to the VCA to V1 volts, obtained from the VARIABLEDC supply.T7 observe the two inputs to the SUMMER. These should be exactly the same asobserved during Task T4. The slope overload should, therefor, beapparent as before.stabilityThere are now three amplifiers in the feedback loop. At the best of times this couldbe a cause for concern - the stability of the whole system could be compromised.Refer to the Appendix to this experiment for further comment.adaptive controlThe VCA is now set up in the feedback loop, but is currently in a passive mode.You are now ready to implement adaptive control of the loop gain by replacing thefixed control voltage V1 with the adaptive control voltage from the modulator.What you will want to observe is the reduction of the length of the period of theslope overload.T8 while watching the length of the slope overload portion of the sawtoothwaveform from the INTEGRATOR, replace the DC voltage from theVARIABLE DC supply to the VCA with the ADAPTIVE CONTROLvoltage from the modulator.Adaptive delta modulationD1- 145

T9 replace the DC control voltage with the ADAPTIVE CONTROL from the SAMPLER.Notice that the slope overload remains as for the conditions of theprevious Task when the control voltage was 4 volt, but that thegranularity at the extrema of the message has not been worsened.To change between adaptive and non-adaptive operation movethe patch cord from the ADAPTIVE CONTROL output socket of theSAMPLER to the preset (V1) output of the VARIABLE DC module.T10 spend some time examining the waveforms at the various interfaces. Asnecessary, replace the ADAPTIVE CONTROL voltage with the manual(DC voltage) control. Don t forget to monitor the ADAPTIVE CONTROLvoltage itself. In other words, make sure you make enoughobservations to appreciate what is happening.T11 use a ‘complex message’, as described in the experiment entitled Deltamodulation, and compare results (by visual inspection of theINTEGRATOR output waveform) with and without the adaptive feedbackoperating.You should now be reasonably confident, from your observations at the modulator(transmitter), that the adaptive feedback control will improve the performance of thesystem as observed at the demodulator (receiver).Thus it might be agreed that the object of the experiment has been achieved.For positive verification, however, it is necessary to build a demodulator and makesome further observations.demodulationIt is essential that you have already completed the experiment entitled Deltademodulation. This introduced methods of noise and distortion measurement, whichare required now.You should now model a delta demodulator, as described in the experiment entitledDelta demodulation.Whilst absolute measurement of signal to noise-plus-distortion ratio (SNDR)measurements are of interest, of greater interest in the present situation is to observethe change to the demodulated waveform which happens when the adaptive feedbackis introduced. This is a qualitative measurement but nonetheless very instructive.The setting up procedure at the demodulator will be somewhat similar to that used atthe modulator.Of interest will be a measurement to resolve the question: is it necessary to make thedemodulator adaptive in the same manner as at the modulator ? Is there a penaltyfor not doing so ?146 - D1Adaptive delta modulation

T12 set up a demodulator. Use a complex message.waveforms under various conditions.Observe recoveredTUTORIAL QUESTIONSQ1 make a positive statement about how your observations at the modulatorconfirmed that the ADAPTIVE CONTROL ‘improved’ the performance ofthe modulator.Adaptive delta modulationD1- 147

APPENDIXloop stabilityYou are working with a feedback loop. At the best of times these can run intoinstability if the loop gain is too high. Some of this instability can be caused by unplanned for phase changes round the loop.SAMPLER outputto SUMMERADAPTIVE CONTROLvoltageDC for meangain controlFigure 4: gain modification using VCAThe presence of two BUFFERamplifiers in cascade does nothelp the situation. These wereplaced there in the non-adaptivemodulator as a convenientmethod of changing the loopgain.Now there is an extra source ofphase change introduced by the VCA, and also a new source of gain adjustment. If itturns out that the gain provided by the two BUFFER amplifiers is near unity it mightbe prudent to remove them.Small gain adjustment could be introduced by the scheme illustrated in Figure 4.This scheme has not been included in any of the Tasks. If you elect to use it, thenmodify the instructions accordingly.148 - D1Adaptive delta modulation

142 - D1 Adaptive delta modulation ADAPTIVE DELTA MODULATION ACHIEVEMENTS: introduction to a variation of the basic delta modulator, which adjusts the step size according to the slope of the signal being sampled PREREQUISITES: completion of the experiments entitled Delta modulation and Delta demodu